// Load data from original text file
static void LoadFromDB()
{
MongoServer server = MongoServer.Create();
MongoDatabase db = server.GetDatabase("pubmed");
MongoCollection<BsonDocument> coll = db.GetCollection<BsonDocument>("patientcare");
MongoCursor cursor = coll.FindAll();
int count = 0;
foreach (BsonDocument article in cursor)
{
BoWModel bowDoc = new BoWModel();
bowDoc.DocID = article["ArticleId"].AsString;
if (!article["Title"].IsBsonNull)
{
SplitWords(article["Title"].AsString, bowDoc);
}
if (!article["MeshHeadings"].IsBsonNull)
{
foreach (BsonValue s in article["MeshHeadings"].AsBsonArray)
{
string s1 = s.AsString.Replace(".", "");
int classLabel = classLabelDict.GetValue(s1);
bowDoc.AddClassLabel(classLabel);
}
}
if (!article["AbstractTexts"].IsBsonNull)
{
foreach (BsonDocument s in article["AbstractTexts"].AsBsonArray)
{
if (!s["Value"].IsBsonNull)
{
SplitWords(s["Value"].AsString, bowDoc);
}
}
}
docModelDB.StoreToDB(bowDoc.StoreToDB());
count++;
if (count % 1000 ==0)
{
Console.WriteLine("Loading {0} records", count);
}
}
coll = null;
db = null;
server.Disconnect();
classLabelDict.StoreToDB();
wordDict.StoreToDB();
}
static void SplitWords(string s, BoWModel docModel)
{
string l = s.ToLower();
Match m;
Regex r = new Regex(@"[a-zA-Z]+[0-9]*");
for (m = r.Match(l); m.Success; m = m.NextMatch())
{
int word = wordDict.GetValue(Detachment.Instance.Detach(m.Value));
docModel.AddWord(word);
}
}
public override DocModel LoadFromDB(BsonDocument bsonDoc)
{
BoWModel docModel = new BoWModel();
return docModel.LoadFromDB(bsonDoc, wordDict);
}